DETTERMAN – THE BEST OF BOTH WORLDS
Detterman (1986, 1987, 1996) claims to have solved the two major (and related) oppositions in the history of intelligence theory:
1. Intelligence as a low-level, global property of all intellectual operations vs. intelligence as a high-level, complex intellectual function – as we have seen already, the first view is advocated by Galton, Spearman and Jensen whereas the second is advocated by Binet, Thurstone and Sternberg.
2. Intelligence as a general ability (again as advocated by Galton, Spearman and Jensen) vs. intelligence as specific abilities (as advocated by Thurstone and Gardner).
Detterman’s solution to these oppositions is to take the ‘best of both worlds’ – general intelligence is real, but rather than being a single ‘ability’, it is better viewed as a high-level property of a complex system composed of multiple intelligences. Detterman argues that the performance of any complex task, including intelligence tests, requires a number of basic abilities. In this scheme, general intelligence represents an average of the processing of several independent components that contribute to the performance of any complex task. This contrast with Spearman’s proposal that there is a single ability common to all tasks and that differences in this single ability between individuals (hypothesized, for example, by Jensen to be represented by differences in speed of processing) give rise to differences in ‘general’ intelligence. Although these conceptions may sound similar, if Spearman is right, we have two important empirical predictions:
1. there should be a single task that correlates as highly with a standard measure of intelligence as measures of intelligence correlate with each other; and
2. if two tasks are correlated with a measure of general intelligence, they should also be correlated with each other.
Detterman considers that both of these predictions are falsified by actual data. Measures of basic cognitive functions in fact have low correlations with each other, and no single basic task correlates with g as highly as Spearman would predict. Detterman cites Guilford (1964), who measured correlations between tests that he believed reflected 150 facets of intelligence. Each test individually correlated with the general measure of IQ derived from the test battery. But a full 17 per cent of 7000 or so correlations between tests were effectively zero. On the other hand, Deary and Stough (1997) have argued that the correlation between inspection time and IQ is high enough to support the prediction that Spearman’s g might be measurable by a single simple task. It is also unclear to what extent in Guilford’s studies the zero correlations between tasks were due to the different reliabilities of the elementary tests used or to the use of participants with a restricted range of abilities. Nevertheless, Detterman has pointed to a new approach to resolving what now very old disputes in the intelligence literature are.
Detterman (1986, 1987, 1996) claims to have solved the two major (and related) oppositions in the history of intelligence theory:
1. Intelligence as a low-level, global property of all intellectual operations vs. intelligence as a high-level, complex intellectual function – as we have seen already, the first view is advocated by Galton, Spearman and Jensen whereas the second is advocated by Binet, Thurstone and Sternberg.
2. Intelligence as a general ability (again as advocated by Galton, Spearman and Jensen) vs. intelligence as specific abilities (as advocated by Thurstone and Gardner).
Detterman’s solution to these oppositions is to take the ‘best of both worlds’ – general intelligence is real, but rather than being a single ‘ability’, it is better viewed as a high-level property of a complex system composed of multiple intelligences. Detterman argues that the performance of any complex task, including intelligence tests, requires a number of basic abilities. In this scheme, general intelligence represents an average of the processing of several independent components that contribute to the performance of any complex task. This contrast with Spearman’s proposal that there is a single ability common to all tasks and that differences in this single ability between individuals (hypothesized, for example, by Jensen to be represented by differences in speed of processing) give rise to differences in ‘general’ intelligence. Although these conceptions may sound similar, if Spearman is right, we have two important empirical predictions:
1. there should be a single task that correlates as highly with a standard measure of intelligence as measures of intelligence correlate with each other; and
2. if two tasks are correlated with a measure of general intelligence, they should also be correlated with each other.
Detterman considers that both of these predictions are falsified by actual data. Measures of basic cognitive functions in fact have low correlations with each other, and no single basic task correlates with g as highly as Spearman would predict. Detterman cites Guilford (1964), who measured correlations between tests that he believed reflected 150 facets of intelligence. Each test individually correlated with the general measure of IQ derived from the test battery. But a full 17 per cent of 7000 or so correlations between tests were effectively zero. On the other hand, Deary and Stough (1997) have argued that the correlation between inspection time and IQ is high enough to support the prediction that Spearman’s g might be measurable by a single simple task. It is also unclear to what extent in Guilford’s studies the zero correlations between tasks were due to the different reliabilities of the elementary tests used or to the use of participants with a restricted range of abilities. Nevertheless, Detterman has pointed to a new approach to resolving what now very old disputes in the intelligence literature are.
No comments:
Post a Comment